Fix STPL_AMER.sh: Add missing SLA parsing and save raw data

Problem

STPL_AMER.sh is missing the SLA normalization step that STPL.sh (EMEA) has. Zendesk returns verbose SLA strings (e.g. 2 hours till breach2hApril 23, 2026 11:26), and without extracting the compact value, the raw strings fail every downstream regex match (^([0-9]+)([mhd])$).

This causes:

• Priority 1: empty - is_within_3_hours() regex never matches, so no New tickets are captured
• Priority 3: empty - is_within_business_hours() regex never matches
• Priority 4: misclassified - tickets that should be P3 fall through to P4
• Sort ordering broken - get_sla_sort_value() returns empty values

Verified with real data

Priority	Expected (with fix)	Actual on main (broken)
P1	712911 (63m), 712930 (2h)	None
P2	712364 (-17h), 707208 (8h)	712364, 707208 (correct)
P3	712958 (8h)	None
P4	712959 (24h)	712959, 712958 (misclassified)

Fix

1. Tidy SLA values after field splitting (matching STPL.sh): extracts compact SLA like 3h, -17h, 63m from verbose Zendesk strings
2. Save raw data to all_data.txt for debugging (matching STPL.sh)
3. Add checkbox formatting to Claude prompt: - [ ] before each numbered ticket, with continuous numbering across priority groups